Progress Report VAULT-SYS 11/20/24

More substructure building and some apologies

Been saving this kit for ten years now. It is going to host the same code I am working on now but likely with a few polyfills and some additional HTML. I am planning to run it on the STM-32 or else something similarly military hardened. Right now it is all I can do to finally release the new, perfected version of this application … then I will work on the fun stuff like this. Will have to wait until I am settled in the mineshaft.

Still aiming for the end of this month or early December. Hoped to have it out before Trump’s inauguration. As prototype, not a release version.

Last version had a lot of fixed sizes of certain critical lists including description lists. As part of my overhaul of the infrastructure, I had to do a lot of editing in code, SQL, resource pack and HTML to accommodate a new model I have come up with. I have taken so much stuff that I used to do with pages of SQL assembly in C code and now replaced it with logic in the resource, including all SQL and barely any processing of REST calls to server. Have plenty of specific RESTs like “get/categorytree” API functions. Many of the REST calls are now just a simple combination of VIEWS in SQLite with serialization to JSON. The fact it is so much simpler is proof to me it is far better written code. The ten pages of sprintf code to make the SQL have changed into just a couple calls to the database.

The idea is that supporting the user in editing lists (as an administrator) to be as big or as small as could possibly be needed without crashing anything.

Category and subcategory of items in inventory used to be hardcoded as lists you could not delete or add to.

This is a screenshot of some REST code/javascript/UI/HTML I was testing on mobile. Same code that runs full desktop also fits in hand on mobile screen. I also wanted to test if the HTML/JQuery performed differently. It didn’t and it looks to be coming along as a universal app that runs on any display. Could be speaking too early so will have to wait and see. Web icons (vector icons for scaling) are just rudimentary while I figure out how to assign them to anything for display.

This is themable and there will be at least ten themes I have now, including my green terminal VDU style. This is a unit test running which adds a random number of unique categories to each node then a random number of subcategories. It’s all assembled on the fly now and delivered via an AJAX call to the server. Supposed to be the station identity info at the top but I turned it off while I was doing unit testing.

This was a series of unit tests adding categories and subcategories then erasing them and refreshing the inventory tree. Sort of works different from previous version. The main discriminator of views into the database coming from clicking categories *(including ALL) to see if the screen refresh worked with a hundred of each and then a thousand of each. No slowdowns, scrolled effortlessly until you get to 3400+ of nodes combined. It is highly unlikely anybody will ever need that many categories and subcategories. All this presents is a data table list into the database of records - you then click in that window to bring up a discrete record. That works with infinite records and zero slowdown. I added ten thousand random entries into inventory, then paged through them and back again. Very fast.

This is all in preparation for people with colossal inventories. The old version would have never kept anywhere near what this new version will. Even the javascript is using intelligent caching of batches of records and is not likely to crash on ten million, although without a finer refinement of your criteria you are liable to be paging for a while looking for the record. No danger now of running out of memory in the browser. Further testing is needed. The further criteria is found at the top of the datatable browser and I am making that loadable with dynamic entries for query searches as well. I have had that running for many years.

I apologize again for running so late, I will likely be late/absent at my own funeral. A lot of these “big engineer” ideas I have were easier to imagine than to implement but as the substructure is tested and works, a lot of the functionality is already working in code. Some of it is at least a decade old and pretty solid. I tested some ancient code I had in there to read from a CUECAT barcoder plugged into a splitter in the keyboard line and it worked right out of the gate! So those of you desperate to finally use your CUECATs in VAULT-SYS the past twenty years will be happy about that I am certain.(!) Still the cheapest barcoder ever sold with the widest recognition of existing code types.

If the company had not tried to install spyware on your computer and gone bankrupt getting sued, we would have never had a chance to own this marvelous device for a measly $10 and change. Beats the hell out of my $9 barcoder … wait a second, I’ve been working on this stupid program so long barcoders that used to cost $50 will now set you back for your loose pocket change. CUECAT, we hardly knew ye.

---

• Regards, Tex